Sökning: onr:"swepub:oai:DiVA.org:uu-526952" >
Network-Based Analy...
Network-Based Analysis of Protein Interactions among Drugs and Adverse Reactions: Identifying Phenotype-Groupings and Key Genes
-
- Ås, Joel (författare)
- Uppsala universitet,Klinisk farmakogenomik och osteoporos,Science for Life Laboratory, SciLifeLab
-
- Eriksson, Niclas, 1978- (författare)
- Uppsala universitet,Uppsala kliniska forskningscentrum (UCR),Klinisk farmakogenomik och osteoporos,Science for Life Laboratory, SciLifeLab,Kardiologi
-
- Hallberg, Pär, 1974- (författare)
- Uppsala universitet,Klinisk farmakogenomik och osteoporos,Science for Life Laboratory, SciLifeLab
-
visa fler...
-
- Wadelius, Mia (författare)
- Uppsala universitet,Klinisk farmakogenomik och osteoporos,Science for Life Laboratory, SciLifeLab
-
visa färre...
-
(creator_code:org_t)
- Engelska.
- Relaterad länk:
-
https://urn.kb.se/re...
Abstract
Ämnesord
Stäng
- Background:Adverse drug reactions (ADRs) present a significant healthcare challenge, leading to morbidity, hospitalizations, and even fatalities. Serious ADRs are in general infrequent, since drugs with a high risk-benefit ratio are rarely approved by the authorities.Genetic factors contribute to serious ADRs, driving pharmacogenomic research to investigate drug-ADR-genetic relationships. These relationships are, however, still largely unstudied due to the scarcity of cases. This scarcity, coupled with the multiple hypothesis problem of genetic studies, poses challenges for these studies. One approach is to group similar ADRs or drugs to bolster sample sizes. However, grouping of drugs and ADRs requires caution to avoid including biologically ill-fitting cases. The objective of our study is to cluster drugs and ADRs based on previous genetic associations and shared protein interactions to propose phenotype groups and genetic targets for investigation.Methods:We developed a Bayesian probability model to substantiate protein-protein interactions across different drugs or ADRs. Subsequently, these proximity values were utilised for spectral clustering to form phenotype-groups. Once obtained, the model was reformulated to rank shared proteins for each cluster.Results:Permutation analysis demonstrated high sensitivity in correctly clustering drugs into therapeutic groups (sensitivity 94-97%) - outperforming other proposed methods - and assigning ADRs to clusters (sensitivity 86%). The model's reformulation enabled the ranking of shared proteins within each cluster, revealing enrichment in KEGG pathways relevant to therapeutic classifications. Discussion:This method successfully replicated known therapeutic drug classifications with high sensitivity, using shared protein interactions among KEGG pathways associated with drug functions. Using the proximity score and spectral clustering we propose phenotype groups and genetic targets for investigations. However, further studies are needed to assess the method's utility for the selection of cases and for target identification in less homogeneous drug-ADR scenarios.
Ämnesord
- MEDICIN OCH HÄLSOVETENSKAP -- Medicinska och farmaceutiska grundvetenskaper -- Medicinsk genetik (hsv//swe)
- MEDICAL AND HEALTH SCIENCES -- Basic Medicine -- Medical Genetics (hsv//eng)
Nyckelord
- Adverse drug reactions; target discovery; study design; protein-protein interaction; statistical modelling; enrichment analysis
Publikations- och innehållstyp
- vet (ämneskategori)
- ovr (ämneskategori)